Mnemonics

xkcd comic Password Strength, showing that the passphrase correct horse battery staple is easier to remember and harder to guess than Tr0ub4dor&3
Image: Randall Munroe https://xkcd.com/936/, CC BY-NC
2026-10

I put the xkcd comic above because of the great tangent the last panel makes: “You’ve already memorized it”. It’s definitely doing something right since I read this comic years ago and was able to search for it with using the query “xkcd horse battery staple”, though I had forgotten about the word “correct”. What that last panel is doing is setting up a memorable image where the man is having a discussion with a horse about a battery that has a staple stuck to it for whatever reason. It’s so weird that it somehow sticks in your memory for years.

Most actual mnemonic techniques work exactly this way, you just need to put a little bit more effort to do it properly, say, to not forget about the word “correct”. You need to imagine a weird or absurd scene where each of the items to be memorized is represented as a clear separate element with minimal risk for confusion with other things. Some kind of ordering may also be memorized, and you might note that the way the last panel is set up suggests the password is “horse battery staple correct” which is incorrect. Setting aside the ordering, the way I would perhaps memorize those four words would be to imagine myself riding a battery-powered horse through gates made of giant metallic staples, while vigorously pulling the reins left and right and yelling “correct the course!”. I’ll also focus in my imagery how the horse is going really fast because the batteries mounted on it are pumping in power and making the steering difficult.

The above is the appropriate level of detail, that if you really visualize it, and then set up an automated “spaced repetition system” (SRS) to quiz you about it at strategica intervals, you will remember for the rest of your life. Typically the SRS system, such as the popular Anki, would quiz you this “card” at intervals like 1 day, 4 days, 10 days, 1 month, 6 months, 3 years, etc. each of which would take you some 10 seconds to answer. And if you did the initial memorization right, it’s likely you would pass all those tests and remember the thing until the next review in 20 years. So beyond setting up the card and doing the vivid imagination, memorizing a single fact like this will ideally take about 1 minute total across three years. It means there is potential to create tens of thousands of cards like this and remember them all without spending all of your time on it (it will take plenty of time though).

I have personally used this kind of mnemonics for a few things: Memorizing the Japanese kanji characters (Heisig method), memorizing long numbers on demand, and also some for some “memory palace” experiments. I will detail these in below.

Major system

At least personally I encounter somewhat commonly situations where it’s useful to quickly memorize 3-8 numbers without using my phone to write them down. When I was a math student it was also handy and fun to memorize constants used in physics. The method I use is a variant of Mnemonic major system which I learned from the magician/mentalist Derren Brown’s book Tricks of the Mind (2007).

The idea of this system is that the numbers to be memorized are first mapped into consonant letters and then you are free to insert vowels to build them into words. When you have the words, then you create vivid imagery to remember them. Then finally when it comes time to recall, the imagery comes to you automatically, you convert it to words, and then you convert the words to numbers again. It sounds like a lot, but really you can do the whole recall in a split second.

Suppose I’m getting on a train and looking at my phone I see my seat number is 329. In my system 3=m, 2=n, and 9=g/k. Now, what word comes to mind when considering vowelless “mng” or “mnk”? For me it’s “monkey”, so what I do is imagine a monkey causing a scene at my seat. Pulling the pillow, jumping around and making monkey noises. It’s important to imagine something vivid, not just an abstract monkey in some empty space. In this imagery the monkey is connected to the train seat itself, which is likely to make it automatically pop up when I enter the train and see some seats. So now I do mentally “monkey” -> “mnk” -> 329, to get my seat number.

To be fair, random number strings don’t always translate to consonants that make nice words. To make this method foolproof you may also need to create strong imagery for each of the individual ten numbers. To remember 329 you might then imagine the pre-created images of 3, 2, and 9 somehow vividly interacting. Or perhaps a mix where the 32 word vividly interacts with the image of 9. Derren Brown’s book and Mnemonic peg system have some extra information on this.

As some final examples, my old mnemonics for physics constants where something along these lines, I haven’t used them in over ten years so I did actually forget them partially:

  • Speed of light: 2.99792 -> nggtgn -> nugget gun -> image about a cannon shooting giant chicken nuggets that reach maximum velocity
  • Planck constant: 6.62607015×10−34 Js -> bbn bst slv mr Js -> baboon beast slave, mr. Ja(m)es -> some image I can’t recall. “beast slave” weren’t the words I used in my original mnemonic I think.

Memory palace

Perhaps the only place I’ve run in to the memory palace concept in the wild is Hannibal LecterHannibal Lecter and to less extent its TV adaptation. In the novel it’s described how Dr. Lecter has memorized basically all of his past life and encounters in this imaginary physical construct that takes the shape of rooms in a castle (he lived in one as a child). Derren Brown and various other sources describe the method, but I will omit that here because personally I have got only limited use of the method.

The core idea is to make use of your spatial intelligence and memory to store “data” in familiar surroundings, such as the past houses and apartments you lived in. If you close your eyes and concentrate, you can probably visualize quite vividly the shapes and details of such places, even if it was many years ago. You may remember positions, shapes, or touch sensation of the furniture to surprisingly high detail. All of that is the good kind of “vivid” memory that can be made to associate with the kind of novel vivid imagery described earlier. So you can attach arbitrary facts to this “palace” of yours, and for example build sequences out of them by imagining yourself moving through the palace.

I haven’t successfully used this technique for memorizing anything useful, but I have employed it as a kind of relaxation technique where the spaces I wander through are also completely imaginary and intended to immerse me in a fantasy world of my own creating. It seems Dr. Lecter also uses the method a lot in this way, as he talks about how he will “live in the memory palace” if he ever gets caught by the police and is put into an uncomfortable cell.

Heisig system

Besides studying for medical school and biochemistry, the overwhelming interest in SRS and mnemonics seems to come language learners, especially those of Japanese and Chinese which require memorization of thousands of complex characters to obtain basic literacy.

The Heisig method is a particular system developed by James Heisig to memorize the writing of thousands of Japanese kanji characters, and later the author modified it also for use with the Chinese hanzi. I personally employed this system many years ago to learn Japanese kanji and definitely got a lot of benefit. The core idea is to break the characters to the repeating patterns and then memorize their combinations using the kind of vivid imagery described above. Frankly, I doubt whether foreign learners can develop sufficient kanji literacy without employing some method of this kind. They don’t have the luxury of life long immersion the Japanese people get growing up.

The books by Heisig offer a very good ordering for memorizing the characters and nicely break down their key parts (similar to “radicals” but more arbitrary, as is necessary). He nicely explains the need for vivid and weird imagery and instructs you to focus on production of the characters with pen and paper. If you only learned to recognize the characters, it would make your knowledge shallow and quick to atrophy.

Having gone through a few passes at the Heisig books, I have just a few suggestions, plausibly already pointed out in Dr. Heisig’s prologues to his books, but worth for me to raise here:


The imagery you choose to employ needs to be vivid for you. For example he sets a distinction of a “big dog” imagined as a St. Bernard, a “small dog” as a chihuahua. But I’m not a dog person so that neither of these is a vivid image for me, so I sometimes struggled when either came up.

Dr. Heisig does suggest you to make up your own imagery for the primitives, and you really need to it for example when as a default he suggests using the retired pro-wrestler Mr. T as symbol for the ever-present “person” primitive. Instead, you really need to conjure your own “person” image that you can imagine in various weird situations, say batman or your imaginary ideal girlfriend.

I also found that many other of Heisig’s suggest imagery were too weak for me and I should have readily took things from my own life to set up these arbitrary primitive meanings. There are primitives that are obvious radicals like “water” that you don’t want to override, but there’s also plenty of repeating primitives that really don’t mean anything in particular, and even if they did, would benefit from being replaced with some vivid imagery you have.